Iterative Amortized Inference
This addresses a key efficiency and accuracy problem in deep latent variable models for researchers and practitioners in machine learning, representing an incremental advancement.
The paper tackles the amortization gap in variational auto-encoders by proposing iterative inference models that optimize inference through repeated gradient encoding, resulting in improved performance over standard models on benchmark image and text datasets.
Inference models are a key component in scaling variational inference to deep latent variable models, most notably as encoder networks in variational auto-encoders (VAEs). By replacing conventional optimization-based inference with a learned model, inference is amortized over data examples and therefore more computationally efficient. However, standard inference models are restricted to direct mappings from data to approximate posterior estimates. The failure of these models to reach fully optimized approximate posterior estimates results in an amortization gap. We aim toward closing this gap by proposing iterative inference models, which learn to perform inference optimization through repeatedly encoding gradients. Our approach generalizes standard inference models in VAEs and provides insight into several empirical findings, including top-down inference techniques. We demonstrate the inference optimization capabilities of iterative inference models and show that they outperform standard inference models on several benchmark data sets of images and text.